2-1. Patient selection
This study was approved by the institutional ethics committee (Yonsei University Health System, Seoul, Korea, 3-2020-0236), and all procedures were conducted in accordance with the ethical standards of the 1964 Helsinki declaration and its later amendments. The requirement for informed consent was waived for this study by Yonsei University Health System as it was based on retrospective, anonymous patient data and did not involve patient intervention or the use of human tissue samples.
We collected data on 449 consecutive patients who underwent MRI/TRUS fusion-targeted biopsies at our institution between January 2017 and September 2020. Indications for prebiopsy MRI with national reimbursement policy were elevated PSA level, identification of palpable nodules during digital rectal examinations, presence of hypoechoic lesions suspected of being PCa lesions, and/or PSA levels >100 ng/mL. We recommended prostate biopsy for patients with PSA levels > 3.0 ng/mL, palpable nodules identified through digital rectal examinations, continuous elevations in PSA levels during follow-up periods, and/or relevant prebiopsy MRI findings.
Of 580 target lesions in the 449 patients (median age: 69.3 years [interquartile range: 62.9–75.3]; median PSA level: 7.02 ng/mL [5.00–11.00]; median PV: 32.8 cm3 [25.7–45.9], median PSA density (PSA level divided by PV): 0.20 ng/mL/cm3 [0.13–0.36]; number of patients diagnosed with PCa: 312 [69.5%]; number of patients with previous prostate biopsy history: 100 [22.3%]; cancer detection rate [CDR] in patients with PI-RADS 3 lesions: 57/180 [31.7%]; CDR in patients with PI-RADS 4 lesions: 156/255 [61.2%]; CDR in patients with PI-RADS 5 lesions: 136/150, [90.7%]), 160 were excluded from the study for the following reasons: (i) for 120 patients, no TRUS images were obtained during MRI/TRUS fusion-targeted biopsies; (ii) for 16 patients, the quality of ultrasound images was poor; and (iii) three patients were diagnosed as having neuroendocrine tumors. Finally, 420 target lesions were included in this study.
2-2. Data collection
Data on patient characteristics, including clinicopathological data such as age, history of prostate biopsy, PV, PSA level, PSA density, MRI findings, and TRUS images, were obtained.
2-3. MRI protocol and imaging analysis
Imaging was performed using a 3-T MRI system (Intera Achieva 3.0 T, Phillips Medical System, Best, the Netherlands), equipped with a six-channel phased-array coil. The prostate MRI protocol involved diffusion-weighted imaging in addition to T2-weighted imaging. Images were acquired through T2-weighted turbo spin-echo MRI in three planes (axial, sagittal, and coronal). MRI datasets were obtained at identical slice locations, with a slice thickness of 3 mm and no intersection gap. Two b-values (0–1400 s/mm2) were used, and diffusion restriction was quantified via apparent diffusion coefficient mapping. Dynamic contrast-enhanced MRI was also performed.
Uro-radiologists denoted suspicious regions of interest on the apparent diffusion coefficient maps examined using the Digital Imaging and Communications in Medicine workstation. The PI-RADS scoring system was used to describe MRI findings [3]. Visible lesions were defined as lesions with PI-RADS scores ≥3.
2-4. MRI-targeted biopsy technique
Prostate biopsies were performed after periprostatic nerve blocks had been administered using Chiba needles. Initially, four MRI-targeted cores biopsies for each targeted lesion were performed, followed by 12-core biopsies. MRI/TRUS fusion targeted biopsies were performed with an MRI/TRUS-fusion-targeted-biopsy protocol using the bk3000 ultrasound system, which involves the use of a side-fire ultrasound probe (BK Medical, Peabody, MA, USA) and an image-based fusion system (BioJet; GeoScan, Lakewood Ranch, FL, USA). All prostate biopsies were performed by an experienced urologist using an 18-gauge, 20-cm disposable core biopsy instrument (Max-Core™; BD, BD Headquarters, NJ, USA).
2-5. Ultrasound image analysis
Ultrasound images of target lesions observed on MRI were stored concomitantly using a picture archiving and communication system (PACS) (GE Healthcare, Barrington, IL, USA). We analyzed the grayscale version of the images using a red/green/blue (RGB) scoring method through a function embedded in the PACS. An average RGB value was obtained from scores at three other randomized points in the most identical slice. Grayscale values were replaced with RGB values on a pixel-by-pixel basis (Y=0.2126*R + 0.7152*G + 0.0722*B) [11]. Two investigators measured the average of three points, and we confirmed there were no differences in grayscale through paired sample t-tests.
2-6. Study objectives
Our primary objective was to investigate the efficacy of imaging analysis by quantifying hypoechoic lesions in MRI-target lesions, demonstrating its diagnostic accuracy, and specifying the quantitation range of hypoechoic lesions for the prediction of PCa and csPC. Our secondary objective was to identify the grayscale range for improving the prediction of PCa and csPC in addition to PI-RADS.
2-7. Statistical analysis
Continuous variables are expressed as medians (interquartile range) or mean ± standard deviation, and categorical variables as number of occurrences and frequency. Pearson’s χ2 test was used for statistical comparisons of continuous and categorical variables. Univariable and multivariable logistic regressions were used. Receiver operating characteristic (ROC) curves and areas under the ROC curves (AUCs) were used to evaluate the performance of standard clinical parameters (including age, previous prostate biopsy history, PSA level, PV, and PI-RADS score) versus the grayscale values in addition to standard clinical parameters for the diagnosis for PCa and csPC (defined as PCa of Gleason grade group ≥2) for target lesions. Pairwise comparisons of ROC curves were conducted to compare the predictive performance of individual and combined parameters. The cut-off value was assessed from the AUCs. These optimal cut-off values were based on predefined values and an analysis performed using the Youden index (sensitivity + specificity – 1). All statistical comparisons were conducted with IBM® SPSS® Statistics 25 (IBM, Armonk, NY, USA) and MedCalc version 11.6 (MedCalc Software Ltd, Acacialaan, Ostend, Belgium). For differences, p<0.05 was considered statistically significant.